Exploring heavy element distribution in soil proles to assess the geochemical background and contamination extent in soils of a metallurgical and industrial area in Kosovo

This paper reports the results of a geochemical study focused on the distribution of heavy elements (As, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, U, Zn) in soil proles, with the aim of dening their geochemical background, contamination threshold, surface and vertical extent of contamination in soils of a metallurgical and industrial area in Kosovo. The geochemical background of heavy elements in soil was assessed by exploratory data analysis of their concentrations in soil proles, and the contamination threshold was used as reference value to quantify their anthropogenic enrichment. Cadmium, Pb, Sb, Zn were the primary soil contaminants, As, Cu, Tl the secondary ones. The main sources of soil contaminants were the Zvecan smelter for Pb, Sb, As, Cu, Tl, and the Trepca industrial complex for Cd and Zn. The highest levels of Pb and Sb contamination were found up to depths between 30 and 60 cm in soil proles within 5 km north and south-east of the Zvecan smelter. Contamination by Pb and Sb decreased with depth in soil proles and affected the whole thickness of soils closer the metallurgical plant. Cadmium and Zn contamination declined with distance from the Trepca industrial complex and decreased with depth, generally extending down to depths of 40–90 cm and 30–70 cm, respectively. Cobalt, Cr, Ni, U were geogenic heavy elements. Anomalous natural concentrations of Co, Cr, Ni were found in soils collected in the northern part of the study area, where the geology consists mainly of ultrabasic and basic magmatic rocks.


Introduction
In environmental geochemistry assessment of the geochemical background (or natural background) is a proper approach to establish whether the concentrations of heavy elements in soil are geogenic due to natural factors and processes (e.g. parent rock, morphology, climate, weathering and erosion), or have been affected by human contributions from industrial plants, vehicle tra c, agriculture and so on. Therefore, de ning the geochemical background is essential for distinguishing the natural and anthropogenic origin of heavy elements in soil, determining the level and extent of contamination, tracing the source of contaminants, and establishing action levels in environmental legislation (Matschullat et al. 2000;Salminen and Gregorauskiene 2000;Baize and Sterckeman 2001;Reimann and Garret 2005;).
In the last 20 years, many studies have assessed the geochemical background of heavy elements (e.g. As, Cd, Cr, Cu, Hg, Ni and Pb) in soil in geo-environmental settings such as industrial and mining sites as well as agricultural and urban areas, by applying geochemical and statistical methods (Chiprés et al. 2009;Pérez-Sirvent et al. 2009;Roca et al. 2012;Hao et al. 2014;Galuska et al. 2016;Reimann et al. 2018;Yotova et al. 2018). Geochemical methods estimate the geochemical background of heavy elements in soil by means of their concentrations in uncontaminated soils in pristine sectors of the study area and in deep layers of soil pro les in contaminated sites, or by using reference values such their average contents in the bulk/upper earth crust and in soil estimated at regional to global scale (Massas et al. 2009;Yang et al. 2009;Lienard and Colinet 2016). Statistical methods assess the geochemical background through the calculation of statistical indices of analytical datasets or techniques of exploratory and spatial data analysis. Exploratory data analysis allows identi cation of populations in datasets (e.g. heavy element concentrations in soil samples) through graphic tools such as histograms, boxplots, cumulative distribution diagrams and probability plots Chiprés et al. 2009;Ander et al. 2013;Reimann and de Caritat 2017).
Assessment of the heavy element geochemical background in soil is a method for de ning the level and extent of contamination in human-impacted areas such as mining, metallurgical and industrial sites. In metallurgical sites atmospheric emissions from non-ferrous ore smelters are a worldwide environmental issue. The smelting plants release heavy elements associated with airborne ne and ultra ne particles that may travel long distances and accumulate in soil by wet and dry deposition (Csavina et al. 2014;Ettler 2016;Lee et al. 2020). In addition, coarser particles blown by wind from mining and ore processing waste disposal sites, usually located near smelters, may be a secondary source of heavy elements (Ettler et al. 2014).
Many studies have revealed severe heavy element contamination in surface soil (topsoil) near smelters, extending a few tens of kilometres and decreasing with distance from the metallurgical site (Sta lov et al. 2010a;Kribek et al. 2010;Ettler et al. 2011;Li et al. 2015;Ghayoraneh and Qishlaqi 2017;Xing et al. 2019). Level and extent of contamination by heavy elements in topsoil around smelters depends mainly on the amount, size, physical properties and chemical/mineralogical composition of the particles released, prevailing winds, landscape features and soil characteristics.
Several authors have investigated the vertical distribution of heavy elements in soils of metallurgical sites, showing a common pattern consisting in a signi cant enrichment in the upper part of the soil pro le and a decline in concentrations with depth (Kabala and Singh 2001;Yuan et al. 2017;Gelly et al. 2019). However, few studies have established the surface and in-depth level and extent of soil contamination by heavy elements around smelting plants in relation to the local geochemical background, usually represented by the element concentrations in selected subsoil layers (Sterckeman et al. 2000;Kribek et al. 2010;Lienard and Colinet 2016;Lee et al. 2020).
In this view, a eld geochemical study was performed to explore the top-to-bottom distribution of heavy elements (As, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, U and Zn) in soil pro les of a metallurgical and industrial area in Kosovo with the main objectives of: i) assessing the local geochemical background and contamination threshold of heavy elements in soil; ii) de ning the surface and vertical level and extent of soil contamination by heavy elements; iii) identifying the primary and secondary soil contaminants and their relationships with the main sources of contamination in the study area.
The study was conducted in an area of northern Kosovo around the town of Kosovska Mitrovica affected by severe and widespread soil contamination by heavy elements such as Cd, Pb, Sb and Zn, mainly due to deposition of particles emitted by the Zvecan smelter and the Trepca industrial complex, and blown by wind from the waste dumps near these plants (Borgna et al. 2009;Sta lov et al. 2010b;Nannoni et al. 2011;Boisa et al. 2013;Šajn et al. 2013;Aliu et al. 2016Aliu et al. , 2017Brewer et al. 2016;Nannoni et al. 2016;Aliu et al. 2017;Prathumratana et al, 2020). Soil contamination by heavy elements is very intense in a large area including the smelter and industrial site, as well as the towns of Kosovska Mitrovica and Zvecan, and extends up to about 20 km from this hotspot, north along the Ibar valley and south-east along the Sitnica valley.
The previous studies in the Kosovska Mitrovica area mostly determined concentrations of Pb and other heavy elements in topsoil (normally up to a depth of 10 cm) to de ne their surface distribution in the entire Mitrovica region or in selected areas. Heavy element enrichment in the soil was quanti ed using reference values from outside the Kosovska Mitrovica area such as the average concentration in European topsoil, Dutch target and intervention values and CLEA (Contaminated Land Exposure Assessment) soil guideline values (Boisa et al. 2013;Šajn et al. 2013;Aliu et al. 2016;Brewer et al. 2016). However, some data is available on the top-to-bottom distribution of heavy elements in soil of this sector of northern Kosovo (Borgna et al. 2009), and a few attempts have been made to assess the local geochemical background of heavy elements in soil. For this purpose, Brewer et al. (2016) applied a statistical method to Pb concentrations in surface and subsurface soils, and Borgna et al. (2009) used As, Cd, Cu, Pb, Sb, Zn concentrations in soil samples collected at depth of 90-100 cm.

Study area
The study area includes a former metallurgical and industrial area in northern Kosovo and extends about 25 km north and south-east of the town of Kosovska Mitrovica along the valleys of the Ibar and Sitnica rivers (Fig. 1). The Sitnica river ows south-east to north-west in a large valley with asymmetrical slopes and meets the Ibar river in the town of Kosovska Mitrovica. The Ibar river ows west to east to Kosovska Mitrovica. Downstream of the con uence with the Sitnica river, it takes a northward direction in a narrow and engraved valley.
The geology of the Kosovska Mitrovica area is complex and consists of Palaeozoic to Quaternary magmatic, metamorphic and sedimentary rocks (Bogdanović et al. 1978). The oldest rocks are Palaeozoic metamorphites (e.g. schists, phyllites, quartzites, metalimestone and metabasites) cropping out east and north-east of Kosovska Mitrovica. The Palaeozoic succession lies under metamorphosed Triassic volcanic-sedimentary formations and limestone. Triassic ultrabasic and basic magmatic rocks (serpentinites, diabases and basalts) crop out in the northern sector of the study area, while Cretaceous clayey-arenaceous-marly ysches characterize the geology of the western sector. Miocene sediments (sandstone, shales, conglomerates, marls and marly limestone) and intermediate volcanites (latites, quartzolatites and their pyroclastic products) are found east and north-east of Kosovska Mitrovica. Large outcrops of Pliocene clayey-sandy sediments and Pleistocene uvio-lacustrine sediments are found in the southern sector of the study area, and recent alluvial deposits border the Sitnica and Ibar rivers.
World-class Pb-Zn mineralizations of the Trepca mining district are found a few kilometres north-east of Kosovska Mitrovica. They are skarn mineralizations hosted by Upper Palaeozoic metalimestone as massive to sub-massive sulphide lenses. The ore assemblage consists mainly of coarse-grained sulphides such as Agbearing galena, sphalerite, pyrite and pyrrhotite.
The Trepca mining district is one of the largest ore districts in Europe consisting of about 40 mines, of which the Stari Trg and Trepca mines were the most important. In the study area metallurgical and industrial plants processed the ore of the Trepca mining district. They included the Zvecan smelter, other secondary ore processing plants and the factories of the Trepca industrial complex. Sulphide ore was initially processed at the Tuneli Pare otation plant and then mainly at the Zvecan smelter about 2 km north of Kosovska Mitrovica (Fig. 1). The Zvecan smelter opened in 1939 and was shut down by the United Nations in 2000 after the Kosovo war. During its operation, various contaminants (e.g. SO 2 and heavy elements) were released to the atmosphere as gas and particles from the stacks of the metallurgical plant. Large quantities of waste from the otation and smelting activities accumulated in the Gornie Polje, Zitkovac and Zarkov Potok dumps close to the Zvecan smelter. The Trepca industrial complex in the eastern suburb of Kosovska Mitrovica included chemical and battery industries and a Zn electrolysis plant (Fig. 1). The waste from these industrial plants went to a large dump nearby. The Trepca industrial complex was closed in 1999 at the end of the Kosovo war.

Soil sampling
Ninety-ve soil samples were collected in nine soil pro les (SP1 to SP9) at different distances from the Zvecan smelter along two transects, hereafter named transects N and SE (Fig. 1). Transect N extends north from the smelter for about 13 km along the Ibar valley; transect SE is about 11 km long and extends south-east from the smelter along the Sitnica valley. The transects N and SE were aligned with the prevailing winds.
The soil sampling sites along the transects N and SE were selected on the basis of the distribution of heavy element concentrations in topsoil reported by Borgna et al. (2009). Soil samples were taken every 10 cm along the soil pro le up to the parent rock down to a depth of 70 to 150 cm, in uncultivated sites within agricultural areas. The parent rock of soil samples consists of the alluvial sediments of the Ibar and Sitnica rivers. The soil sampling was performed using Edelman augers.

Soil treatment and analysis
In the laboratory, the soil samples were manually disaggregated, dried at + 40°C and then sieved using a 2 mm mesh sieve. Soil particles less than 2 mm were homogenized by quartering and pulverization. About 250 mg of pulverized soil sample was solubilized by acid digestion using a mixture of ultrapure reagents consisting of 2 mL HF, 2 mL HNO 3 and 1 mL H 2 O 2 (US EPA 1996). Solubilization was performed in Te on bombs using a Milestone ETHOS 900 microwave digester.
Concentrations of As, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, U and Zn were determined by inductively coupled plasma mass spectrometry (ICP-MS) using a Perkin Elmer NexION 350 spectrometer. Analytical accuracy was assessed by analysis of the standard reference materials SRM 2709 (San Joaquin Soil) and SRM 2710 (Montana Soil) from the National Institute of Standards and Technology (NIST). Analytical precision was calculated by ve replicate analyses of each soil sample.

Geochemical index
The single pollution index (PI) is a geochemical index widely used for quantifying heavy element contamination of soil due to human contributions (Hu et al. 2013;Kowalska et al. 2018). A prerequisite for its calculation is knowledge of the geochemical background of heavy elements in soil. In this study the single pollution index was calculated as the ratio of heavy element concentration in the soil sample to its contamination threshold, namely the upper limit of its geochemical background in soil. The following scale of contamination was used: PI < 1 contamination absent; 1 < PI < 2 low contamination; 2 < PI < 3 moderate contamination; 3 < PI < 5 strong contamination; PI > 5 very strong contamination.

Statistical analysis
The quantile-quantile (Q-Q) plot was used as the graphic technique of exploratory data analysis to identify populations in the whole dataset of As, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, U and Zn concentrations in soil samples. The concentrations of heavy elements (whole dataset and single populations) were processed to calculate the main statistical indices: min and max values, mean, median, standard deviation and coe cient of variation (CV).
The Spearman non-parametric correlation test (p-level < 0.001) was applied to de ne correlations between concentrations of heavy elements in soil samples.

Results And Discussion
Geochemical background of heavy elements As shown in Table 1, the concentrations of As, Cd, Co, Cr, Cu, Ni, Pb, Sb and Zn in soil samples collected from soil pro les in the Kosovska Mitrovica area were highly variable (CV from 0.54 to 2.06), while those of Tl and U were more uniform (CV≈0.25). Exploratory data analysis suggested that concentrations of Cd, Pb, Sb and Zn in soil samples belonged to four main populations (A, B, C, D), As, Cu and Tl to three populations (A, B, C), Co, Cr and Ni to two populations (A, B), and U to one population (A; Table 1). This partitioning of heavy element concentrations in population coupled with literature data on their contents in uncontaminated soils and in rocks similar to those cropping out in the Ibar and Sitnica valleys, was applied to assess the geochemical background of As, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, U and Zn in the soils of the study area. Table 1 Statistics of As, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, U and Zn concentrations (in mg/kg) in all soil samples (all data) and populations A, B, C, D (SD = standard deviation; CV = coe cient of variation) On this basis, the concentrations of As, Cd, Cu, Pb, Sb, Tl and Zn belonging to population A were taken as representative of their geochemical background in soils formed by the alluvial sediments of the Ibar and Sitnica rivers ( Table 2). In other words, the minimum and maximum concentration values in population A identify the geogenic variability of these heavy elements in soils of the study area due to natural factors and processes (e.g. mineralogical and geochemical characteristics of parent rocks, morphology, climate and weathering).
Consequently, the upper limit of the geochemical background of As, Cd, Cu, Pb, Sb, Tl and Zn in soil corresponds to the respective contamination threshold, and concentrations of these heavy elements in populations B, C, D identify variable levels of contamination.  (Gao et al. 1998;Reimann and de Caritat 1998).

Heavy element distribution in soil pro les
To investigate the vertical distribution of As, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, U and Zn, heavy elements and soil pro les were categorized and grouped according to the contamination level of topsoil (0-10 cm) de ned by the value of the single pollution index (PI).
The rst category of heavy elements included Cd, Pb, Sb and Zn: their concentrations in topsoil were always higher than the corresponding contamination threshold and indicated a prevalence of a strong to very strong contamination. Based on the contamination level of Cd, Pb, Sb and Zn in topsoil, two groups of soil pro les were distinguished: Group 1, soil pro les with low to moderate contamination (1 < PI < 3); Group 2, soil pro les with strong to very strong contamination (PI > 3).
In soil pro les of Group 1 concentrations decreased progressively up to a depth of 30 cm for Pb and Sb, 20-40 cm for Zn and 30-50 cm for Cd (Fig. 2). At greater depths, heavy element contents were quite uniform and within the corresponding geochemical background.
Soil pro les of Group 2 showed a similar distribution of Cd, Pb, Sb and Zn characterized by a vertical decrease in concentrations that involved a wider thickness of pro le or even the whole pro le (Fig. 2).
Cadmium and Zn exhibited progressive and sharp declines in content down to depths of 50-90 cm and especially in the rst 50 cm of soil pro le (Fig. 2). Below, the concentrations of these heavy elements were homogeneous and geogenic. Only in soil pro le SP6, located close to the Trepca industrial complex, Zn contents were higher than the respective contamination threshold down to the bottom of the pro le.
Concentrations of Pb and Sb decreased progressively down to the bottom of the soil pro les SP2, SP3, SP4, SP5 and SP6, although the decline was particularly marked in the upper 50 cm (Fig. 2). In these soil pro les contamination by Pb and Sb affected the whole soil thickness. In soil pro les SP7, SP8 and SP9 from the southern sector of the study area, Pb and Sb concentrations decreased down to depths of 50-70 cm and 30-40 cm, respectively. At greater depths they were quite uniform and within the corresponding geochemical background.
Enrichment of Cd, Pb, Sb and Zn in the upper layer and progressive decline in their concentrations with depth were also found in the Kosovska Mitrovica area by Borgna et al. (2009)  The results of this study recall the vertical distribution of Cd, Pb, Sb and Zn in soils contaminated by metallurgical activities. Steckerman et al. (2000) found that Cd, Pb and Zn concentrations decreased abruptly in the upper part of soil pro les and contamination was mostly in the rst 20-30 cm. They also found that migration of these heavy elements increased with their concentrations in topsoil. Similarly, Cd, Pb, Sb and Zn were signi cantly enriched in the upper 20 cm of soils near a Pb-Sb smelter in southwest China (Yuan et al. 2017). Moreover, the distribution of Pb concentrations in the strongly to very strongly contaminated soil pro les of the present study is consistent with the results of Gelly et al. (2019), who found that Pb contamination involved the whole thickness of soil pro les close to a Pb-Ag smelter. These authors estimated the human contribution to be more than 90% of the total element content in topsoil and 65-78% in subsoil.
In metallurgical sites accumulation of Cd, Pb, Sb and Zn in the upper layer of soil pro les is mainly the result of: i) deposition of heavy element-bearing airborne particles emitted by smelters or blown by wind from mining dumps; ii) adsorption and precipitation/coprecipitation of the heavy element aliquot mobilized in soil by chemical alteration of the smelter-derived particles. Adsorption of heavy elements mainly occurs on the surface of organic matter (e.g. humic substances), Fe-Mn oxyhydroxides and clay minerals, precipitation as own minerals such as carbonates, oxides/hydroxides, sulphates and phosphates, and coprecipitation with Fe-Mn oxyhydroxide colloids (Wilson et al. 2010;Yuan et al. 2017;Lee et al. 2020). These redistribution processes mainly depend on the alteration of particles emitted by smelters, soil physico-chemical properties (e.g. pH and cation exchange capacity), content and characteristics of reactive soil constituents (e.g. clay minerals, Fe-Mn oxyhydroxides and organic matter) as well as element behaviour in adsorption and precipitation/coprecipitation.
Soil enrichment of Pb and Zn in surface soils due to precipitation of own minerals such as carbonates, sulphates and phosphates and/or coprecipitation in minerals such as calcite, was mainly found in contaminated neutral to alkaline calcareous soils (Jacquat et al. 2008;Liang et al. 2014). In this regard, Nannoni et al. (2011) detected cerussite (PbCO 3 ), hydrocerussite (Pb 3 (CO 3 ) 2 (OH) 2 ) and minor anglesite (PbSO 4 ) in highly Pb-contaminated topsoils formed by the alluvial sediments of the Ibar and Sitnica rivers. These soils had neutral to sub-alkaline pH (6.7-7.6) and a variable carbonate content from < 1-14.8%.
The downward migration of Cd, Pb, Sb, Zn and other heavy elements in soils contaminated by smelters is ruled by several factors including level of contamination, physical characteristics and composition of smelter-derived particles, soil properties, surface reactivity of soil constituents, element speciation and transport in soil solution, behaviour in adsorption and precipitation/coprecipitation reactions, land use and climate (Fernandez et al. 2008;Tyszka et al. 2016;Lee et al. 2020). Cadmium, Pb, Sb and Zn move down through soil solution both in dissolved forms as free ions and ionic and non-ionic complexes, and in particulate forms adsorbed by colloids (e.g. humic substances, Fe-Mn oxyhydroxides and clay minerals), as well as included as main or trace constituents in smelterderived ne and ultra ne mineral particles (Citeau et al. 2003;van Oort et al. 2006;Gelly et al. 2019). Studies on vertical movement in soil pro les showed that Pb and Sb were less mobile than Cd and Zn, which were predominantly in dissolved forms in soil solution (Denaix et al. 2001;Clemente et al. 2008). Nevertheless, the downward migration of Pb in soil may be quite fast from 0.3 to 1.36 cm/year (Ettler et al. 2004(Ettler et al. , 2011Prapaipong et al. 2008;Lee et al. 2020). Since the Zvecan smelter started working in 1939, these vertical movement rates sustain the depth of Pb contamination in soils of this study, which normally extended down to 100 cm.
The second category of heavy elements included As, Cu and Tl: their concentrations in topsoil indicated that the upper part of soil pro les was uncontaminated to moderately contaminated (Fig. 3). Soil pro les SP3, SP4 and SP5 were the only exception as they showed strong to very strong As contamination in topsoil. Based on the contamination level of these heavy elements in topsoil, two groups of soil pro les were distinguished: Group 1, soil pro les with topsoil uncontaminated by As, Cu and Tl (PI < 1); Group 2, soil pro les with topsoil affected by low to moderate contamination by Cu and Tl (1 < PI < 3) and low to very strong contamination by As (1 > PI > 5).
In soil pro les of Group 1, the natural concentrations of As, Cu and Tl decreased slightly in the upper 40 cm or had a uniform distribution with depth (Fig. 3). In contaminated soil pro les of Group 2, As and Cu showed a decrease in concentration with depth and reached geogenic contents at depths between 30 and 90 cm and 20 and 60 cm, respectively (Fig. 3). A rapid decrease in As concentrations occurred in the upper 50 cm of strongly to very strongly contaminated soil pro les SP3, SP4 and SP5. Below 50 cm, As concentrations were uniform and slightly higher than the contamination threshold or within the geochemical background. As a rule, Tl concentrations declined in the upper 20-40 cm of soil pro les of Group 2 and were slightly above the contamination threshold in the whole thickness of soil pro les SP5, SP7, SP8 and SP9.
The vertical distribution of As, Cu and Tl in soils of this study was similar to that found in soil pro les of areas contaminated by smelter emissions. In particular, the pattern of Cu concentrations in the soil pro les of this study was comparable with the results of Brewer et al. (2016) in soils of the Kosovska Mitrovica area down to a depth of 50 cm. Ettler et al. (2014) showed that Cu accumulated, mostly bound to organic matter, in the upper part of soil pro les near a Cu smelter in the northern sector of the Zambian Copperbelt. Similarly, Tl was enriched in the upper layer, especially in the organic horizon, of soil pro les from a Zn smelting site in southern Poland and its concentrations declined with depth (Vaněk et al. 2013). These ndings suggest a role of insoluble organic matter in the retention and accumulation of Cu and Tl in surface soils. Nevertheless, migration of Tl down to the bottom of soil pro les was recognized by Vaněk et al. (2013), probably due also to element transport in particulate form adsorbed by colloids of clay minerals and Mn oxides.
The third category included the geogenic heavy elements (Co, Cr, Ni and U): their concentrations in soils from pro les were constantly within the corresponding geochemical background.
In soil pro les SP1 and SP2 the anomalous natural concentrations of Co, Cr and Ni showed no signi cant variation with depth, while in the other pro les the vertical distribution pattern was variable. In fact, the normal natural concentrations of Co, Cr and Ni had both a homogenous distribution and an irregular variation in depth, as well as a slight increase down to depths of 40-70 cm. In soil pro les natural concentrations of U decreased down to 30-40 cm or were rather homogeneous with depth.
The surface and vertical distribution of heavy elements considered to be soil contaminants in the Kosovska Mitrovica area (As, Cd, Cu, Pb, Sb, Tl and Zn) was also investigated in relation to the distance from the Zvecan smelter along the transects N and SE, focusing on topsoil (0-10 cm), the 50-60 cm layer and the bottom of the soil pro le.
The highest concentrations of Pb (1003-4813 mg/kg) and Sb (13-54.9 mg/kg) were found in topsoil of pro les SP3, SP4, SP5 and SP6 up to about 5 km from the Zvecan smelter (Fig. 4). At greater distances from the metallurgical plant, Pb and Sb concentrations in topsoil progressively decreased toward north along the Ibar valley, whereas they were quite uniform toward south-east along the Sitnica valley. This distribution pattern of Pb and Sb in topsoil was the same at the 50-60 cm layer, whereas it was slight for Pb and absent for Sb in the bottom of soil pro les.
As shown in Fig. 4, the peak of Cd and Zn concentrations (23.6 and 3674 mg/kg, respectively) was recorded in the topsoil of pro le SP6 near the Trepca industrial complex. This nding indicates that the industrial activities at the Trepca complex in uenced the distribution of these heavy elements in soils of the Kosovska Mitrovica area. Furthermore, the very high concentrations of Cd and Zn measured by Brewer et al. (2016) in the waste piles of the Trepca industrial complex (especially waste of the Zn electrolysis plant) suggest that dispersal of these materials by wind was the main source of contamination. A marked decrease in Cd and Zn concentrations was found in topsoil of pro les SP7 and SP5, located about 2 km south-east and 5 km north-west of the Trepca industrial complex (7.5 and 10.7 mg/kg for Cd; 1317 and 846 mg/kg for Zn). At greater distances from the industrial plants, the contents of these heavy elements declined further, reaching concentrations below 1.5 mg/kg for Cd and 300 mg/kg for Zn in topsoil of pro les at the limit of the transects N and SE. Conversely, Cd and Zn contents were rather uniform at the 50-60 cm layer and the bottom of all soil pro les, though slightly higher concentrations were measured in soil pro le SP6.
Similarly to Pb and Sb, the highest concentrations of Cu and Tl (80.2-134 mg/kg and 0.84-1.3 mg/kg, respectively) were recorded in topsoil of pro les within 5 km of the Zvecan smelter ( Fig. 4). At the 50-60 cm layer cm and the bottom of the soil pro les, Cu and Tl did not show any variation trend in relation to distance from the contamination sources.
In line with the spatial distribution in surface soils (0-5 cm) of the Kosovska Mitrovica area (Sta lov et al. 2010b), As showed the highest concentration (388 mg/kg) in topsoil of pro le SP3 from the Ibar valley, about 4 km north of the Zvecan smelter (Fig. 4). Arsenic concentrations decreased sharply north and south of this site and were higher in topsoil of pro les from the Sitnica valley (29.6-80.6 mg/kg). Unlike in topsoil, As contents were quite uniform in the 50-60 cm layer (16.4-40.6 mg/kg) and at the bottom (14.2-45.4 mg/kg) of soil pro les along the transects N and SE.
Level and extent of heavy element contamination The upper part of soil pro les showed very strong contamination by Pb and Sb, except pro les SP1 and SP2 located 13 and 11.5 km north of the Zvecan smelter. The soil thickness most contaminated by Pb (PIs = 5.2-81.6) and Sb (PIs = 5.1-38.4) was found up to depths between 30 and 60 cm in soil pro les within 5 km north and south-east of the metallurgical plant (SP3, SP4, SP5, SP6), and between 10 and 20 cm in soil pro les 6 to 11 km south-east of this contamination source (SP7, SP8, SP9). Moreover, PI values suggest that Pb and Sb contamination: i) decreased progressively with depth in all soil pro les; ii) involved the whole thickness of soil pro les closer to the Zvecan smelter (SP2, SP3, SP4, SP5, SP6), where the bottom soil showed low to moderate contamination; iii) reached its highest level in soil pro le SP5 about 1 km from the metallurgical plant.
Most soil pro les showed very strong contamination by Cd (PIs = 5.3-38.1) and Zn (PIs = 5.4-32.8) down to depths of 20-40 cm and 10-50 cm, respectively. Cd and Zn contamination peaked in the soil pro le closest the Trepca industrial complex (SP6) and decreased with distance from this site. Low to moderate contamination characterized the upper part of soil pro les most distant from the metallurgical and industrial plants (SP1, SP2, SP9). In all soil pro les Cd and Zn contamination declined with depth, usually reaching depths of 40-90 cm and 30-70 cm, respectively.
Arsenic contamination was found in soil pro les down to a depth of 90 cm and was low to moderate (PI < 3), except in pro les SP3, SP4 and SP5 in which strong to very strong enrichments (PIs = 3.3-15) were found up to depths between 20 and 40 cm.
Low to moderate Cu contamination was recorded in the rst 40 cm of soil pro les within 5 km of the Zvecan smelter. Low Tl enrichment was detected in soil pro les near the metallurgical plant (SP4, SP5) and industrial site (SP6, SP7, SP8) as well as close to the town of Vucitrn (SP9).
Level and extent of heavy element contamination in the study area suggest that Cd, Pb, Sb and Zn were the primary soil contaminants, while As, Cu and Tl were secondary contaminants. The main sources of contamination were the Zvecan smelter for Pb, Sb, As, Cu and Tl, and the Trepca industrial complex for Cd and Zn. These associations of the heavy element contaminants were con rmed by signi cant positive correlations between the concentrations of As, Cd, Cu, Pb, Sb and Zn in soil samples (Table 3). The most signi cant correlations (r = 0.77-0.87; p < 0.001) concerned the primary contaminants, among which Pb-Sb (r = 0.87) and Zn-Cd (r = 0.79) showed the highest correlation coe cients. The secondary contaminants, As and Cu, were signi cantly correlated between them (r = 0.59) and with Cd, Pb, Sb and Zn, but the correlation coe cients (r = 0.50-0.70) were lower than those between the primary contaminants. The only exception was As-Sb (r = 0.81), presumably due to the similar behaviour of these heavy elements in soil. No signi cant correlations were found between Tl and the other soil contaminants. The results of correlation analysis also sustain the natural origin of Co, Cr, Ni and U in the soils as there were no signi cant positive relationships between concentrations of these geogenic heavy elements and the primary and secondary contaminants, except for the correlation U-Tl (r = 0.67). Conversely, signi cant positive correlations with high coe cient values (r = 0.78-0.95) were found between the concentrations of Co, Cr and Ni, suggesting that these heavy elements had a similar distribution in parent rock and behaviour in soil.

Conclusions
The geochemical background and contamination threshold of heavy elements such as As, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, U and Zn, in soils of a metallurgical and industrial area in Kosovo were determined by exploratory data analysis of their concentrations in top-to-bottom soil samples from pro les. The contamination threshold of these heavy elements was used as reference value to quantify the anthropogenic enrichment of heavy elements in soil and to de ne the surface and vertical level and extent of soil contamination.
Cadmium, Pb, Sb and Zn were the primary soil contaminants as these heavy elements showed the highest levels and widest extent of contamination, while As, Cu and Tl were secondary contaminants as their contamination levels were generally low to moderate. The main sources of soil contaminants were the Zvecan smelter for Pb, Sb, As, Cu and Tl, and the Trepca industrial complex for Cd and Zn.
Lead and Sb showed the highest levels of contamination in the upper part (between 30 and 60 cm) of soil pro les within 5 km north and south-east of the Zvecan smelter. Contamination by these heavy elements decreased with depth in all soil pro les and affected the whole thickness of soils closer the metallurgical plant.
Contamination by Cd and Zn declined with distance from the Trepca industrial complex and decreased with depth, usually extending down to depths of 40-90 cm and 30-70 cm, respectively.
Among the heavy elements analysed, Co, Cr, Ni and U were geogenic as their concentrations in soil were constantly within the corresponding local geochemical background. Anomalous natural concentrations of Co, Cr and Ni were found in soils from the northern part of the study area, where the geology consists mainly of ultrabasic and basic magmatic rocks.
The results of this study highlight that exploration of the top-to-bottom distribution of heavy elements in soil pro les is an effective geochemical approach to assess their geochemical background in soil of contaminated areas.
Declarations Funding Not applicable.

Con icts of interest/Competing interests
The authors declare that they have no con icts of interest and competing interests.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.